Hot and cold numbers do not predict the draw. Here is what they actually do.

What are hot and cold numbers, and what does "overdue" mean?

Hot means drawn most often over a window of past draws. Cold means least often. Overdue is a cold number with a story attached, the story being that it is owed an appearance.

The first two are descriptions, and they are accurate. The third is a prediction, and it fails.

Every frequency chart here counts the last 90 draws. Nothing in that record carries forward, because the machine has no memory of it. The operator says so on its own results pages: every number has an equal chance of being drawn, every time.^[1]

The belief that a number becomes due has a formal name. It is the gambler's fallacy, one of the most studied biases in gambling research.^[3]

Do hot and cold numbers work? No, and here is the test we ran

No, and we checked with the site's own data rather than an argument.

Each of the four Australian games carries a rolling window of 90 draws. We split every window in half. From the older 45 we took the hottest numbers and the coldest, as many of each as the game draws, then counted what they did across the newer 45, which the selection had never seen.

If hot numbers carry momentum, the hot set should run ahead. If cold numbers are due, the cold set should catch up.

Across all four games the hot set appeared 188 times. The cold set appeared 189. If past frequency means nothing at all, the expected figure is 182 apiece. One standard deviation is 11.5 appearances. Both results land inside it.^[5]

Note what that does not say. The cold set did not do better by finishing one appearance higher. A gap of one, where ordinary variation is eleven, is not a finding. Both sets turned up at their fair rate.

The most drawn Powerball numbers in Australia, and why the gap means nothing

Across the 90 Powerball draws from 28 November 2024 to 13 August 2026, number 15 came up 25 times and number 31 came up 10.^[5]

Here is the arithmetic that defuses that gap. Ninety draws of seven main numbers is 630 numbers across a pool of 35, so the average number should appear 18 times. That is a small sample. Some land well above 18 and some well below, and the distance between the extremes is wide by construction. A window this size with no visible hot and cold numbers would be the surprising result.^[5]

Those two figures are a snapshot of the 90-draw window on the date given below, and they will shift every time a new draw is added. Number 15 will not hold the top of that list for long, and nothing about it earned the position while it held it.

Are hot and cold numbers real? The chart is real, the prediction is not

The frequencies are genuine counts of draws that happened, checkable against the operator's published results.

What fails is the step after the count. Reading a record as a forecast requires the balls to know their own history, and they do not. One camp reads the chart and plays the hot numbers. The other reads the same chart and plays the cold. That is the clearest sign the chart is not doing the work.

So the answer is yes and no, in that order. The numbers are real. The signal people read into them is not.

The bit no other Australian site will tell you

Frequency charts cannot predict the draw. They can, to a degree, predict the crowd, and that is where the only real effect here lives.

A study of the Maryland numbers game found that money staked on a number falls sharply after it is drawn, then takes months to recover.^[2] Players were avoiding a number because it had just come up. Later work across far more entries found the same, and other research found players lean toward lines that look representative.^[4]

Both halves of what follows matter. Most coverage keeps one.

Choosing numbers other players avoid does not improve your chance of winning. Nothing does. The odds are fixed by the format of the game, and every combination has the same chance every time. What it changes is the number of ways a prize is divided in the event that it is won. Australian lottery divisions are shared among everyone holding the winning line, so an unpopular line that wins is split fewer ways than a popular one.

The effect is real, small, and impossible to measure from outside: no operator publishes how many tickets held each combination. It is also the only place in this subject where a player's choice changes their outcome.

What this does to the numbers on this site

We publish frequency charts and run generator modes that lean hot and cold. Both are here on purpose, and neither is a claim about the next draw.

The window is 90 draws. A lottery is random, so the longer the window, the more every count converges on the flat average. Ninety is long enough to be a genuine record and short enough that the shape stays visible. That is a choice about a readable chart, not a claim about a correct sample size.

The generator has four modes. She'll Be Right samples uniformly, with no weighting at all. Scorcher leans toward numbers that have come up. Brass Monkeys leans the other way. Straight Down the Line sits between them.

The lean is gentle by design, tuned so that it never becomes exclusion. At the setting we ship, a game's hottest number comes out at roughly 2.4 times its fair rate and the coldest at roughly 0.35 times. Every number stays in play, in every mode.

That matters for the reason above. A mode that froze out cold numbers would steer every user toward the same shrinking handful, which is the prize-sharing problem in miniature. The lean changes which lines come out, not what any line is worth.

The straight version

Hot and cold numbers do not work as a prediction, and no amount of data will make them. Every draw is independent. That is not a caveat on the strategy. It is the whole of it.

The charts are still worth looking at. They are an accurate picture of what a game has done lately, and they are why this site counts anything. Just be clear about what you are holding: a photograph of the past, not a window into Tuesday night.

And if you are going to play the cold numbers anyway, play them for the honest reason. Not because they are due. Because fewer people are on them, which is about the other players, not the balls.

Correct as at 20 August 2026. Frequency figures are counted here from this site's own rolling 90-draw windows rather than quoted, and they change with every new draw. Australian results are published by The Lott. We are not the operator and we do not sell entries. No technique changes the odds of a random draw. Lottery play is for adults 18+.

References

1. The Lott, Oz Lotto results and number frequencies ("each number has an equal chance of being drawn"): thelott.com/oz-lotto/results ↩
2. Clotfelter, C.T. & Cook, P.J. (1993), "The 'Gambler's Fallacy' in Lottery Play," Management Science 39(12): scholars.duke.edu/publication/859676 ↩
3. "A longitudinal analysis of the hot hand and gambler's fallacy biases," Judgment and Decision Making (2023), Cambridge University Press: cambridge.org ↩
4. "The cross-sectional 'Gambler's Fallacy': Set representativeness in lottery number choices," Journal of Economic Behavior & Organization (2014): sciencedirect.com ↩
5. Every frequency figure here is computed from this site's own rolling 90-draw windows for the four Australian games: 360 draws, 2,340 main numbers. The split test takes the top and bottom numbers of the older 45 draws, as many of each as the game draws, ties broken in favour of the lower ball number, then counts them across the newer 45. Breaking ties the other way gives 183 and 193, the same conclusion. The expected count and standard deviation are the four games' hypergeometric means and variances summed, checked against a 20,000 trial simulation. ↩ ↩ ↩ ↩